get_best_motif.py

#!/usr/bin/env python

'''
parses arguments using argparse
@return args list of all parameters
'''
def parse_arguments():
    parser = argparse.ArgumentParser(description='A script to convert from GLAM2 output to MEME-format and parsing only the [num] first motifs from file to the output.')
    parser.add_argument("meme", help="Path to 'meme' file generated by GLAM2")
    parser.add_argument("output", help="Output file")
    parser.add_argument("num", help="Number of motifs parsed from file")
    parser.add_argument("id", help="Cluster ID")
    args = parser.parse_args()
    return args

'''
The script has two functions:
    1. Writing lines of file till certain line (MOTIF + [num])
    2. Converting GLAM2 output to minimal meme-format
@params meme STRING Path to 'meme' file generated from Meme suite
@parmas output STRING Output file
@params num INT Number of motifs parsed from file

@author René Wiegandt
@contact rene.wiegandt(at)mpi-bn.mpg.de
'''
def main():

    args = parse_arguments()
    out = open(args.output, "w+")

    '''
    Create pattern where script should stop writing
    For Example:
    If num == 3, which means that you want the first/best 3 Motifs, the script
    should stop writing lines to output if loop reaches line 'MOTIF  4'
    '''
    number = int(args.num) + 1
    break_header = "MOTIF  " + str(number)

    # Get cluster_id
    cluster_id = args.id

    # Pattern for motif header
    pattern = re.compile("^MOTIF\s{2}(\d)+")
    # Init count
    count = 0

    with open(args.meme) as f:
        for line in f:
            if break_header in line:
                # line matches breaking_header, e.g. 'MOTIF  4'
                break
            else:
                ## do not write [count] lines after each header -> needed for meme-format
                if count > 0:
                    count-=1
                    continue
                if pattern.match(line):
                    # if line is a motif header
                    count = 2
                    out.write(line.strip('\n').replace("  "," ") + " Cluster_" + cluster_id + '\n')
                ##
                else:
                    out.write(line)


if __name__ == "__main__":
    import argparse
    import re
    main()